Scientists at the University of Illinois Urbana-Champaign have developed a new framework to predict how different species within microbiomes interact to create unique compositions. Microbiomes are diverse communities of microbes that play a crucial role in shaping human health. The researchers aim to understand these interactions to harness the potential of microbial therapeutics, which could be used to treat diseases without relying on antibiotics.

To overcome the challenge of assessing interactions between each microbe species across diverse environments, the research team created a model that could predict outcomes of microbial communities based on the initial microbes present. This model provides a “landscape interaction value” that characterizes the effect each microbe has on another’s abundance and the overall outcome of the microbiome.

The researchers found that most of the interactions between microbes had minimal impact on the outcome of the microbiome, with only a select few playing a crucial role. They used compressive sensing to extract more information from sparse datasets. The model was trained using existing microbiome datasets and validated through real-world experiments.

The study’s findings suggest that microbial landscape interactions are often near zero, indicating that the outcomes of microbiomes may be simpler than expected. The researchers believe that understanding the sparsity of these interactions could provide insights into how microbiomes are assembled and how species interact.

The team plans to investigate why so many microbial landscape interactions are near zero and explore larger datasets to see if the patterns they found hold true. They also aim to fine-tune the model for studying specific microbiomes of interest and use it in personalized medicine to predict the risk of certain pathogens establishing within patients’ microbiomes.

The research discussed above was published in PNAS.

References

• Arya, S., George, A. B., & O’Dwyer, J. P. (2023). Sparsity of higher-order landscape interactions enables learning and prediction for microbiomes. Proceedings of the National Academy of Sciences, 120(48), e2307313120. https://doi.org/10.1073/pnas.2307313120
• Lawson, S. & University of Illinois at Urbana-Champaign. (2023, November 30). New model allows for learning and prediction of microbial interactions. Phys.Org; University of Illinois at Urbana-Champaign. https://phys.org/news/2023-11-microbial-interactions.html